A comparison of three CT systems using a phantom, a test object filled with known iodine solutions, found that photon-counting CT (PCCT) generally detected and quantified iodine at lower concentrations than dual-source CT (DSCT) and split-filter CT (SFCT) across most tested protocols. But the lower end of the range where measurements stayed linear was similar across systems. The largest reported reductions in the detection threshold reached 68% for brain scans and 45% for abdominal scans. No human participants were involved.
The clearest gap appeared at low dose
The clearest numerical gap came in low-dose abdominal imaging at 3 mGy, using a medium reconstruction setting. For a 3 mm region of interest, a selected area measured in the image, the limit of detection, or LoD, was 0.78 mg/mL with PCCT, compared with 1.28 mg/mL with SFCT and 1.42 mg/mL with DSCT. For a 10 mm region, the corresponding values were 0.16, 0.26 and 0.57 mg/mL.
Across the low-dose protocols, higher dose was associated with lower detection and quantification thresholds for both region sizes, and the gains were generally greater for the smaller region. The effect of higher iterative-reconstruction levels diminished as dose increased. The pattern was not uniform for every system or metric.
A controlled test of iodine signal
Researchers tested three Siemens CT systems under four head and abdominal low- and high-dose settings. The abdominal dose settings were 3 and 13 mGy, while the head settings were 15 and 50 mGy. Each system and protocol was repeated 20 times.
The phantom was a 15 by 20 cm PMMA cylinder containing iodine-water solutions at eight set concentrations, from 2 mg/mL down to 0 mg/mL, including 0.03125 mg/mL and a zero-iodine sample. A 25 by 35 cm annulus, 15 cm thick, simulated abdominal attenuation. The phantom omitted tissue heterogeneity and anatomical complexity.
The analysis built calibration curves by averaging signal from a central region across selected slices and the 20 repeated acquisitions, then used linear least-squares regression to convert image values into iodine concentration. It estimated three thresholds: the limit of detection (LoD), the limit of quantification (LoQ), and the lower limit of the linear interval (LLLI). PCCT's LoD and LoQ were lower across most protocols, while LLLI was similar across systems.
For LoQ, the team used concentrations above the LLLI and interpolated a precision profile at a 20% coefficient of variation, a measure of relative measurement variability. Some values above the 2.0 mg/mL maximum calibration concentration were marked as extrapolated.
What the numbers do and do not mean
When the researchers compared signal response and noise, PCCT had the highest calibration slope and the lowest noise. DSCT and SFCT had comparable noise, while DSCT had a higher signal response.
To estimate uncertainty, the slice-wise analysis selected every second slice, used circular 3 mm or 10 mm regions of interest, and averaged slice-specific metrics. Results were reported as plus or minus two standard errors. Because selected slices may have remained correlated, the reported uncertainty could be underestimated.
The comparison was limited to one vendor, vendor-recommended protocols, two ROI sizes and a limited range of dose levels. The phantom omitted tissue heterogeneity and anatomical complexity. The work was a phantom study with no human participants.
Karolinska University Hospital supported the work through the internal Research and Development fund of its Department of Nuclear Medicine and Medical Physics, and Karolinska Institute provided open-access funding. One author, PN, was an employee of Siemens Healthineers Sweden and reportedly did not control the data; the other authors reported no relevant conflicts or industry support. The analyzed datasets are available from the corresponding author upon reasonable request.
Paper data and sources
Original title: Evaluation of iodine detection and quantification limits in photon-counting CT—a phantom study
Authors: Titternes R, Poludniowski G, Brismar T et al.
Journal/Repository: European radiology experimental
Status: Peer-reviewed
First online: 2026-10-12
DOI: Not available
Original paper · Full text